HyperSU: Corpus-Driven Semantic-Unit Hypergraph for Retrieval-Augmented Generation

• URL: http://arxiv.org/abs/2606.28351v1
• Date: Wed, 03 Jun 2026 14:59:47 GMT
• Title: HyperSU: Corpus-Driven Semantic-Unit Hypergraph for Retrieval-Augmented Generation
• Abstract summary: HyperSU is a novel hypergraph-based RAG framework featuring semantic-unit hyperedges and clue-guided bidirectional retrieval.<n> Experiments show that HyperSU consistently improves answer accuracy over standard, graph-based, and hypergraph-based RAG baselines.
• Score: 5.684949359748965
• License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
• Abstract: Recent Hypergraph-based retrieval-augmented generation (HyperRAG) methods use hyperedges to connect multiple entities simultaneously, enabling more efficient multi-entity evidence organization than pairwise graph structures. However, existing HyperRAG methods often rely on LLM-generated summaries to construct hyperedges, which can introduce hallucinations while also incurring high indexing costs. In addition, during retrieval, existing methods typically rely on either one-hop neighbor expansion or PageRank diffusion. The former may miss useful multi-hop evidence, while the latter can suffer from uncontrolled propagation over excessive hub nodes, leading to semantic drift and noisy reasoning chains. To address these challenges, we propose HyperSU, a novel hypergraph-based RAG framework featuring semantic-unit hyperedges and clue-guided bidirectional retrieval. During construction, HyperSU formulates hyperedge construction as an entity-aware minimum-description-length (MDL) optimization problem, inducing source-grounded semantic-unit hyperedges that balance sentence-level semantic coherence and entity compactness. It then constructs a hypergraph by modeling each semantic unit as a hyperedge over its co-mentioned entities. During retrieval, HyperSU performs clue-guided bidirectional expansion over the semantic-unit hypergraph, enabling both multi-hop evidence discovery and answer-aware noise reduction. Experiments show that HyperSU consistently improves answer accuracy over standard, graph-based, and hypergraph-based RAG baselines, achieving up to a 14.7% relative accuracy improvement on GraphRAG-Bench, with larger gains on reasoning-intensive tasks.

Related papers

• Hypergraph as Language [49.35033483821496]
  We propose a hypergraph-native alignment framework for large language models.<n>For hypergraph structures, existing methods often fail to preserve the native semantics that multiple objects are jointly connected by the same high-order relation.<n>We show that Hyper-Align significantly outperforms existing methods across in-domain and zero-shot evaluations.
  arXiv  Detail & Related papers  (2026-05-21T01:09:48Z)
• HELP: HyperNode Expansion and Logical Path-Guided Evidence Localization for Accurate and Efficient GraphRAG [53.30561659838455]
  Large Language Models (LLMs) often struggle with inherent knowledge boundaries and hallucinations.<n>Retrieval-Augmented Generation (RAG) frequently overlooks structural interdependencies essential for multi-hop reasoning.<n>Help achieves competitive performance across multiple simple and multi-hop QA benchmarks and up to a 28.8$times$ speedup over leading Graph-based RAG baselines.
  arXiv  Detail & Related papers  (2026-02-24T14:05:29Z)
• BHyGNN+: Unsupervised Representation Learning for Heterophilic Hypergraphs [50.161252392272324]
  We introduce BHyGNN+, a self-supervised learning framework for representation learning on heterophilic hypergraphs.<n>By contrasting augmented views of a hypergraph against its dual using cosine similarity, our framework captures essential structural patterns in a fully unsupervised manner.<n>Our results validate the effectiveness of leveraging hypergraph duality for self-supervised learning.
  arXiv  Detail & Related papers  (2026-02-16T16:55:37Z)
• HyperRAG: Reasoning N-ary Facts over Hypergraphs for Retrieval Augmented Generation [22.189132611244105]
  HyperRAG is a RAG framework tailored for n-ary hypergraphs.<n>It learns structural-semantic reasoning over n-ary facts to construct query-conditioned relational chains.<n>It achieves the highest answer accuracy overall, with average gains of 2.95% in MRR and 1.23% in Hits@10 over the strongest baseline.
  arXiv  Detail & Related papers  (2026-02-16T05:15:55Z)
• HyperSearch: Prediction of New Hyperedges through Unconstrained yet Efficient Search [32.06048071728742]
  Given a hypergraph, hyperedge prediction aims to identify hyperedges that are either missing or likely to form in the future.<n>Existing approaches rely on either sampling to obtain constrained candidate sets or ungrounded assumptions on hypergraph structure to select promising hyperedges.<n>We propose HyperSearch, a search-based algorithm for hyperedge prediction that efficiently evaluates unconstrained candidate sets.
  arXiv  Detail & Related papers  (2025-10-20T04:55:40Z)
• Directional Sheaf Hypergraph Networks: Unifying Learning on Directed and Undirected Hypergraphs [14.396507207496507]
  We introduce Directional Sheaf Hypergraph Networks (DSHN), a framework integrating sheaf theory with a principled treatment of asymmetric relations within a hypergraph.<n>DSHN achieves relative accuracy gains from 2% up to 20%, showing how a principled treatment of directionality in hypergraphs, combined with the expressive power of sheaves, can substantially improve performance.
  arXiv  Detail & Related papers  (2025-10-06T11:46:53Z)
• HyperGCT: A Dynamic Hyper-GNN-Learned Geometric Constraint for 3D Registration [60.01977041900338]
  HyperGCT is a flexible dynamic Hyper-GNN-learned geometric ConstrainT.<n>It mines robust geometric constraints from dynamic hypergraphs for 3D registration.<n>Experiments on 3DMatch, 3DLoMatch, KITTI-LC, and ETH show that HyperGCT achieves state-of-the-art performance.
  arXiv  Detail & Related papers  (2025-03-04T02:05:43Z)
• Hyperbolic Hypergraph Neural Networks for Multi-Relational Knowledge Hypergraph Representation [19.14148664582895]
  We propose the Hyperbolic Hypergraph Neural Network (H2GNN), whose essential component is the hyper-star message passing.<n>We compare H2GNN with 15 baselines on knowledge hypergraphs, and it outperforms state-of-the-art approaches in both node classification and link prediction tasks.
  arXiv  Detail & Related papers  (2024-12-11T12:03:33Z)
• Hypergraph Transformer for Semi-Supervised Classification [50.92027313775934]
  We propose a novel hypergraph learning framework, HyperGraph Transformer (HyperGT) HyperGT uses a Transformer-based neural network architecture to effectively consider global correlations among all nodes and hyperedges. It achieves comprehensive hypergraph representation learning by effectively incorporating global interactions while preserving local connectivity patterns.
  arXiv  Detail & Related papers  (2023-12-18T17:50:52Z)
• Augmentations in Hypergraph Contrastive Learning: Fabricated and Generative [126.0985540285981]
  We apply the contrastive learning approach from images/graphs (we refer to it as HyperGCL) to improve generalizability of hypergraph neural networks. We fabricate two schemes to augment hyperedges with higher-order relations encoded, and adopt three augmentation strategies from graph-structured data. We propose a hypergraph generative model to generate augmented views, and then an end-to-end differentiable pipeline to jointly learn hypergraph augmentations and model parameters.
  arXiv  Detail & Related papers  (2022-10-07T20:12:20Z)
• Adaptive Neural Message Passing for Inductive Learning on Hypergraphs [21.606287447052757]
  We present HyperMSG, a novel hypergraph learning framework. It adapts to the data and task by learning an attention weight associated with each node's degree centrality. It is robust and outperforms state-of-the-art hypergraph learning methods on a wide range of tasks and datasets.
  arXiv  Detail & Related papers  (2021-09-22T12:24:02Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.